Stimulating the vagus nerve after practice helped new skills stick in mice

Four 15-minute eye-tracking sessions a day were enough to train the mice, and a small electrical pulse to the vagus nerve after each one was enough to change how well they remembered it. At Tohoku University’s Graduate School of Life Sciences, Junyu Chen, Yoko Ikoma and Professor Ko Matsui implanted a cuff electrode on the left cervical vagus nerve of mice and found that stimulation delivered after practice, not during it, strengthened learning on later days. The paper appeared in iScience on August 25, 2026.

It is a mouse study with an invasive electrode, and it was run on a single type of skill. Matsui’s group at Tohoku frames the result around timing: the stimulation came after the session, and that placement is what the experiment tests.

Cuff electrodes and the optokinetic task

The stimulation itself was delivered through a small cuff electrode wrapped around the left cervical vagus nerve, the branch that runs through the neck carrying signals between the body and the brain. Matsui stresses that stimulation was delivered only after training, so any benefit cannot come from stimulation during the task itself.

The skill was horizontal optokinetic response learning, known as HOKR, a cerebellum-dependent task in which a mouse’s eye movements are trained to follow a moving visual pattern, loosely like tracking a passing train. Tohoku’s Graduate School of Life Sciences release describes the task as dependent on the cerebellum and lists the paper under DOI 10.1016/j.isci.2026.117413, titled “Vagal nerve stimulation induces vascular oscillations and enhances long-term learning.”

As described by PsyPost, mice completed four 15-minute training sessions a day, and some received electrical stimulation afterward while controls received none. The team watched pupils and heart rate to confirm that the stimulation was reaching its target.

The effect appeared on a delay. Stimulation did not improve performance during training itself. Mice that had received it showed stronger eye-tracking performance by days 2 to 5, with higher-intensity pulses producing benefits by day 2 and lower-intensity pulses reaching similar gains by day 5. According to News-Medical’s account, the gain was in long-term learning on subsequent days, not in same-session performance.

Blood vessels that pulse near the cerebellar flocculus

To look for a mechanism, the team used fiber photometry to track blood volume in the cerebellar flocculus, the region that controls eye movements. Each round of stimulation triggered a two-phase vascular response: a brief fall in blood volume followed by a delayed rise, repeating as a rhythm.

Mice that showed the largest vascular oscillations also showed the biggest learning gains by day 5. Matsui described the idea behind the work in the Tohoku account: “VNS may open a hidden window of opportunity for enhanced learning by making the brain environment more receptive to long-lasting change.” Chen, the lead author, frames the broader point as the body shaping the brain more strongly than is usually assumed, with metabolic tuning of the brain as one route by which latent learning capacity could be released.

The link between the rhythm and the learning is a correlation. In a first-person account on the Springer Nature community site, the authors state that the work does not prove vascular oscillations cause better learning, and that they did not directly measure ATP, lactate, pyruvate or oxygen metabolism.

The two phases of the vascular response also give a timeline. The brief fall in blood volume comes first and the delayed rise follows, and the rhythm repeats across stimulation. The team tied the size of that rhythm to the later learning curve, and Matsui’s “window of opportunity” language is the group’s way of saying the stimulation may leave the brain better prepared to consolidate what was practiced earlier, rather than to boost practice while it is happening.

Mice, not consumer devices

Matsui was direct about how far the result travels. “This was a mouse study using invasive cervical VNS, so the findings should not be directly generalized to humans or to non-invasive consumer stimulation devices,” he said, as quoted by PsyPost. The distinction matters because a surgically placed cuff on the nerve and a non-invasive consumer device are different interventions, and only the first was tested in this experiment.

The task is also narrow, which limits how far the result reaches. An eye-movement reflex trained in a cerebellar circuit is a long way from learning a piano passage or a golf swing, and the public accounts of the paper give neither the number of mice nor the stimulation frequency, so the size of the effect cannot be expressed as a percentage improvement.

The data establish a timing rule. In this model, a pulse after the session did the job that a pulse during it did not, and larger blood-vessel rhythms went with larger gains. The unresolved item is the one the authors flag themselves: translating the approach to humans depends on later optimization studies, including whether any version of it could work without surgery.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview